A review of three major fungal diseases of Coffea arabica L. in the rainforests of Ethiopia and progress in breeding for resistance in Kenya

Journal of Advanced Research (2011) 2, 109–120

Cairo University

Journal of Advanced Research

REVIEW ARTICLE

A review of three major fungal diseases of Coffea arabica
L. in the rainforests of Ethiopia and progress in breeding
for resistance in Kenya
Holger Hindorf
a
b

a,*

, Chrispine O. Omondi

b

University of Bonn, Phytomedizin, Nussallee 9, D-53115 Bonn, Germany
Coffee Research Foundation, POB 4, Ruiru, Kenya

Received 13 August 2009; revised 30 December 2009; accepted 25 February 2010
Available online 2 October 2010

KEYWORDS
Coffee;
Diseases;
Breeding;

Resistance;
Ethiopia;
Kenya

Abstract In a review of their own research the authors summarize incidences and distributions of
the most important fungal diseases in Ethiopia and progress in breeding for resistance. Ethiopia, as
the centre of origin for Coffea arabica, hosts a large diversity of germplasm. The incidences of
diseases are based on observations in the montane rainforests of the southeast (Harenna) and southwest (Bonga, Berhane-Kontir, Yayu) of Ethiopia. Major diseases are Coffee Leaf Rust (CLR),
Hemileia vastatrix; Coffee Berry Disease (CBD), Colletotrichum kahawae and Coffee Wilt Disease
(CWD), Gibberella xylarioides (Fusarium xylarioides). CLR incidences in Ethiopia were present in
all regions with highs between January and March and lows between June and October. CBD was
present mostly in Bonga (40.0%) and Yayu (26.3%), but less frequent in Harenna (18.6%) and
Berhane-Kontir (6.0%). CWD as a recently developed disease in Arabica coffee could be detected
ranging from 2.4% in Berhane-Kontir to 16.9% in Yayu. CLR has been a serious constraint in all
production countries since it became prominent in Ceylon in the late 19th century after leaf infection defoliation affects plants. CBD was first observed in Kenya in 1922. The disease is currently
confined to the African continent in all countries that grow Arabica coffee. In the mid-1990s in
the Democratic Republic of Congo, Uganda and Tanzania a resurgence of CWD in Robusta coffee
and in Ethiopia in Arabica coffee occurred. Over the last 40 years breeding activities have been car-

* Corresponding author. Tel.: +49 228 732438; fax: +49 228 732442.
E-mail address: (H. Hindorf).
2090-1232 ª 2010 Cairo University. Production and hosting by
Elsevier B.V. All rights reserved.
Peer review under responsibility of Cairo University.
doi:10.1016/j.jare.2010.08.006

Production and hosting by Elsevier


110

H. Hindorf and C.O. Omondi
ried out to combat CLR, CBD and CWD. Breeding for resistance against CLR in Arabica coffee
has successfully utilized single or combinations of major genes designated as SH genes. Major gene
resistance has also been deployed in breeding for resistance against CBD, whereas in the case of
CWD, selections of tolerant Arabica accessions are being pursued from local landraces in Ethiopia.
ª 2010 Cairo University. Production and hosting by Elsevier B.V. All rights reserved.

and Yayu) of Ethiopia. Details of the field sites are shown in
Table 1.

Introduction
The following review of coffee diseases comprises first a
description of three major fungal pathogens: Coffee Leaf Rust
(CLR), Hemileia vastatrix, Coffee Berry Disease (CBD),
Colletotrichum kahawae and Coffee Wilt Disease (CWD),
Gibberella xylarioides (Fusarium xylarioides) including the historical occurrence, distribution, symptomatology, biology of
the pathogens and their economic importance [1]. Control
measures of such immense disease agents are essential; therefore, in the second part of the review sustainable efforts in
breeding for resistance are described. The presented data are
based on the experimental experiences and activities of both
authors and their working teams in Ethiopia and Kenya.
The host, Coffea arabica L.
The genus Coffea is endemic to Africa and a number of species
are described in West, Central and East Africa. Due to disease
constraints and other factors such as yield, quality and growth
habits, only two species are nowadays commercially grown
worldwide, namely C. canephora (Robusta) in lowlands and
C. arabica (Arabica) in highlands. Arabica coffee is grown in
altitude ranges between 1400 and 1800 m and is cultivated

under shade. This species originated from the province of Kaffa
in Ethiopia and was distributed by Yemen traders all over the
world during the 15th century. Today, in a few remaining rainforests of southwest and southeast Ethiopia, coffee grows as an
understory shrub in a large diversity of shade trees, shrubs and
annual plants and has maintained its own genetic diversity as a
natural gene-bank. But even this natural resource is not free of
diseases. It continues, however, to survive all attacks by pathogens and pests in a unique way under natural conditions. Therefore the description and occurrence of diseases will concentrate
on experiences in the montane rainforests of Ethiopia.

The pathogens
The disease frequency of indigenous coffee in the four major
rainforest areas in 2005 is taken to represent the situation in
general during the investigation period of 2003–2008 (Fig. 1).
Coffee Leaf Rust (Plate 1A), H. vastatrix
Coffee Leaf Rust (CLR) is one of the most important diseases
of C. arabica in the world. It devastated Arabica coffee plantations in Ceylon at the end of the 19th century and was
responsible for its replacement with tea plantations. Despite
effective fungicides and resistant varieties developed to control
rust, yield reductions of 20% or more in various countries are
still caused by the pathogen [2]. In Brazil, losses have been estimated to be about 30% and an annual loss of about 4500 tons
of coffee was estimated in Kenya in the 1960s. The pathogen
prefers a temperature range of 20–28 °C, needs a leaf wetness
period only during spore germination and penetrates with
the germination hyphae into the stomata of the host. The fungus tolerates longer seasons without rainfall and spores are
wind-borne, only attacking leaves and needs no other host

CBD

%

CWD

100

CLR
80
60
40
20

Field sites in Ethiopia

0
Harenna

Investigations of the occurrence of diseases were carried out in
four different rainforest regions of the southeast (Harenna in
the Bale Mountains) and southwest (Bonga, Berhane-Kontir

Table 1

BerhaneKontir

Yayu

Disease incidence in indigenous coffee 2005 [24].

Field sites of indigenous coffee of southeast and southwest Ethiopia.

Code, habitat and plots

I. Harenna 1, 2, 3
II. Bonga 1, 2, 3
III. Berhane-Kontir 2, 3
IV. Yayu 1, 2
a

Fig. 1

Bonga

Latitude (N)
0

6°30
7°300
7°050
8°200

CV = coefficient of variation [%].

Longitude (E)
0

39°45
36°300
35°300
35°500

Altitude (m a.s.l.)

1580–1610
1750
1200–1320
1530–1600

Annual precipitation
Mean (mm)

CVa (%)

850
1700
2100
1800

26
16
13
11

(20)
(44)
(25)
(32)


Breeding for resistance to coffee diseases

111

Plate 1 Symptoms of fungal diseases of coffee. (A) Coffee Leaf Rust: on seedlings, older leaves, upper and lower site of the leaf and
hyperparasitized by Verticillium lecanii. (B) Coffee Berry Disease: on green and mummified berries, mycelium colour on Malt-Extract
Agar. (C) Coffee Wilt Disease: dead tree, brownish vascular system on stem, conidia of the imperfect stage Fusarium xylarioides.

for completing the life cycle. Due to the fact that coffee is a
perennial host with green leaves all through the year, the
pathogen produces only uredinio- and teliospores with basidiospores. Coffee grown in lower altitudes is more predisposed
to the disease and suffers more attacks. A heavy infestation of
leaves not only reduces the assimilation area but also results in
a complete defoliation diminishing the next year’s crop
tremendously.
CLR was first reported in Ethiopia in 1934 [3], but the disease had existed for a long time in other countries without causing epidemics or eradications of certain varieties of C. arabica.
The long-term coexistence of coffee and rust coupled with the
high genetic diversity of coffee populations and a high level
of horizontal resistance might have kept the rust at low levels
[4]. Other factors such as the low average productivity associated with shade and the existence of biological agents such as
the hyperparasite Verticillium lecanii, were also believed to play
an important role in maintaining CLR at low levels.

A large number of urediniospore samples were collected in
the Ethiopian rainforests and identification was carried out
during 2003/04 in the Institute of Botany, Tu¨bingen University
[5]. Measurements of urediniospores of CLR from the indigenous coffee population revealed detailed data with typical sizes
for the species of H. vastatrix and had spore dimensions
between 29.7l · 18.9 (minimum) and 34.5l · 23.7 (maximum).
These spore sizes could be compared with those identified in
highly susceptible Ethiopian selections such as Arba, Guga
and Harrar and others from Indonesia and Colombia. The
results showed that measurements were to a large extent identical and confirmed the presence of the species H. vastatrix

(Table 2). The identification proof of the species H. vastatrix
by morphological characteristics was assisted by scanning electron microscopic photos of rust sori and urediniospores [5]. A
typical sorus extruding from a stoma on the lower side of the
leaves had 15–25 lemon-shaped one-celled urediniospores
(Fig. 2).


112
Table 2

H. Hindorf and C.O. Omondi
Sizes of urediniospores of Hemileia spp.

Location

Coll. date

Length

Width

Variations

I. Harenna 1
II. Bonga 1
II. Bonga 2
II. Bonga 2
II. Bonga 2
II. Bonga 2
II. Bonga 2

II. Bonga 3
II. Bonga 3
III. Berhane-Kontir 2
III. Berhane-Kontir 2
III. Berhane-Kontir 3
IV. Yayu 1
IV. Yayu 1
IV. Yayu 2
IV. Yayu 2

8.2004
5.2004
11.2003
11.2003
11.2003
11.2003
11.2003
1.2004
5.2004
1.2004
5.2004
1.2004
11.2003
5.2004
11.2003
5.2004

33.7
32.7
30.9

30.5
30.1
30.0
30.9
30.3
31.8
33.2
29.7
32.1
30.4
30.3
31.2
34.4

22.1
23.7
20.4
21.2
19.8
20.5
20.4
18.9
23.3
19.5
21.9
19.9
20.4
22.7
19.8
21.3

31–36 · 21–23
29–36 · 21–26
30–33 · 20–22
29–32 · 20–23
28–31 · 18–21
28–32 · 20–22
30–33 · 20–22
29–33 · 18–20
27–37 · 20–26
30–36 · 18–21
26–33 · 17–25
30–34 · 19–21
29–32 · 20–22
27–34 · 20–25
30–33 · 19–20
31–38 · 19–23

Mean

2003/04

31.61

21.19

26–40 · 17–26

Urediniosorus and urediniospores of Hemileia vastatrix.

There was little emphasis on race-typing of Ethiopian rust
samples until the beginning of the 1980s and the 1990s, when
the Institute of Biodiversity Conservation (IBC, formerly
gene-bank) included coffee in their conservation system.
Wondimu et al. [6] observed that race III was the most frequent in forest coffee and race II in other areas. Other races
were I, X and XV. In 2005 the first race-typing of CLR collections of indigenous coffee was carried out at the Centre of Coffee Leaf Rust Research (CIFC) in Oeiras, Portugal using their
differentials (Varzea, personal communication). In this recent
study the race specification identified race II at Berhane-Kontir
and race III and X in Bonga with corresponding virulence
genes v 1, 4 and 5 [7].
CLR assessments in the rainforests of Ethiopia revealed
its presence in all fields differing in incidence with time (season) and location. A significantly (P < 0.001) high rust incidence of 31.1% was recorded, for instance, in 2008 at Yayu,
followed by Berhane-Kontir (21.4%) and Bonga (7.9%) in
forest coffee populations. Rust incidences were consistently
highest in Yayu, lower in Berhane-Kontir and lowest in
Bonga forests during all seasons. The occurrence of rust
in the forest coffee populations varied significantly from season to season (P < 0.001). Higher rust incidences were
found in January (29.6%) and April (22.7%), while lower

Harenna

Bonga

Berhane Kontir

Yayu

4.5

Disease Index DI

Fig. 2

3.5
2.5
1.5
0.5
-0.5

Sep Jan May Sep Jan May Sep Jan May Sep Jan
03 04 04 04 05 05 05 06 06 06 07
Period Sep 2003 - Mar 2007

Fig. 3 CLR severity during 2003 and 2007 in indigenous coffee
populations of Ethiopia [24].

incidences were observed in July (13.9%) and October
(14.3%).
Comparing rust occurrence during the complete period of
the surveys from 2003 to 2007 a slight increase of the disease
could be observed in the wild coffee population (Fig. 3).


Breeding for resistance to coffee diseases
The effect of shade on the occurrence of CLR could be
shown in nursery experiments at the Jimma Agricultural
Research Centre (JARC). All young coffee trees grown under
the shade were infected more seriously with rust than in the
non-shaded sites. Comparing coffee from the different forest
regions, the material from Bonga seemed to be more tolerant

to rust than others [7].
Coffee Berry Disease (Plate 1B), C. kahawae
CBD was first detected in 1922 in Kenya around Mt. Elgon,
west of the Rift Valley [8,9]. Soon after detecting the disease,
losses of up to 75% were reported. This brought the coffee
cultivation west of the Rift Valley to a near end and tea plantations became predominant in the region. The dry Rift Valley
stopped the spread to the major coffee-growing areas in the
highlands of the Central Province for a long time. In 1951 a
first appearance of CBD east of the Rift was reported by Rayner [10].
At the beginning, the disease was related to the fungus C.
coffeanum described from Brazil by Noack [11] causing leaf
spots on Arabica coffee. But the new disease in Kenya produced anthracnose-like symptoms on green berries. Rayner
[10] called the pathogen C. coffeanum var. virulans to differentiate between leaf and berry symptoms. Morphological and
pathogenicity research by several authors from the 1960s to
1990s finally resulted in the name C. kahawae, representing
the Kiswahili word for coffee in the species name [12]. Prior
to that time, the pathogen was called either CBD-strain [13]
or C. coffeanum Noack sensu Hindorf [14,15]. Intensive investigations on the Colletotrichum population in coffee were carried out by Hindorf [16–18] and three distinct species occurring
in association with CBD on coffee berries were described as (1)
the CBD-causing species C. coffeanum growing with black colour on artificial Malt-Extract Agar, (2) C. acutatum with pink
colour in vitro and (3) C. gloeosporioides producing symptoms
only on ripe berries as the so-called late blight and a perfect
stage of Glomerella cingulata [17].
From Kenya the disease spread to Angola in 1930, Zaire in
1937, Cameroon between 1955 and 1957, Uganda in 1959,
Tanzania in 1964, Ethiopia in 1971 and Malawi in 1985
[19,20]. Until now the disease has been restricted to East, Central and South African coffee-growing regions. In Ethiopia the
disease occurred much later than in neighbouring Kenya. After
its first appearance in Sidamo and the first report by Mogk
[21], the disease spread very quickly to nearly all growing coffee provinces until 1978 and caused remarkable losses. In the

most restricted province of Hararghe the disease occurred only
after 1985 and the coffee crop started being replaced by Chat,
Catha edulis [22].
The pathogen can infect all organs of the host: flower buds,
leaves, fruits and the maturing bark. Infection takes place
either early during flower bud formation causing some losses
in flowers or remains latent in the inflorescence until the berries
start to expand in growth [23]. The outbreak of the disease
with visible symptoms occurs during the expanding stage of
berry development, producing sunken, black, anthracnose-like
lesions on the green pulp. High moisture or pulp wetness favours the production of conidia in black acervuli appearing
in concentric rings and exuding pink masses of one-celled,
straight or slightly curved hyaline conidia. The conidia are

113
splash-borne or distributed by insects, coffee pickers or other
vectors, but never by wind due to a sticky constellation in
the pink masses. In the absence of buds and berries the pathogen survives in the maturing bark of secondary branches.
The pathogen never attacks mature coffee beans; it remains
in the pulp. The losses occur during early infestation by
destroying the beans or by preventing proper wet and dry processing since the pulp cannot be removed completely, causing
so-called ‘‘stinkers’’ in the crop and reducing the quality. An
intensive progress of the disease in the expanding stage of
the berry development finally produces mummified berries
with no economic value at all.
Information concerning the incidence of CBD in the Ethiopian forest coffee regions of Harenna, Bonga, Berhane-Kontir
and Yayu is presented in Table 3 [24]. Assessments of the incidence (infected trees per locality) and severity (infestation of
single trees) were scored visually. The CBD occurrence
depended mostly on altitude ranges; higher sites were more
frequently infected than lower sites due to more favourable

climatic conditions for the pathogen (Fig. 4).
The pathogenicity of CBD isolates was not only tested on
detached berries in the laboratory but also on seedlings in
the greenhouse to investigate the diversity of coffee grown
under natural conditions. Seedlings from seeds collected in
Harenna, for instance, produced in the lower site 2 incidence
rates of 23.3% and proved to be as similarly resistant/tolerant
as resistant cultivars such as cv. 754 and 741. In contrast, on
the higher site 3 of the same region only one tree with a lower
intensity of 27.3% berries infected by the pathogen of CBD
was found; all the other nine trees were highly susceptible.
Due to the fact that CBD was present in the surroundings
of the Bonga and Yayu sites it was decided to carry out attached berry tests directly in the field, a well-documented
method of testing CBD resistance. The pathogen isolates used
for infection tests were collected from local field sites (Table 3).
The infection tests on attached berries in the field sites of Bonga and Yayu produced a large diversity in susceptibility. Infection rates at Bonga varied in 2004 between 0% and 47.0% and
in 2005 between 7.9% and 81.5%. Coffee trees were less susceptible at Bonga than at Yayu.
Coffee Wilt Disease (Plate 1C), G. xylarioides (F. xylarioides)
Coffee Wilt Disease (tracheomycosis) is a vascular disease
caused by the fungal pathogen, G. xylarioides (F. xylarioides)
and results in a total death of the infected coffee trees. The disease has been a serious problem to the production of Robusta
coffee in DR Congo and Uganda since the 1990s killing hundreds of trees.
The first appearance on Arabica coffee in Ethiopia was reported in 1958 by Lejeune [25] and a diagnostic confirmation
was provided by Kranz and Mogk [26]. The disease occurred
first on some large scale state farms near Gera [27]. Detailed
morphological studies and pathogenicity tests were carried
out in a Ph.D. thesis by Adugna [28], who compared isolates
from Arabica and Robusta coffee. In seedling tests it was
proved that isolates from Arabica sources could only infect
C. arabica and isolates from Robusta sources only infected

C. canephora [28–31]. Therefore it was suggested, that the coffee wilt population should be classified into two formae speciales [32]: G. xylarioides f. sp. abyssiniae (F. xylarioides f. sp.


114
Table 3

H. Hindorf and C.O. Omondi
Incidence and severity of CBD in the forest coffee areas of Ethiopia.

Locality

Sample site

Isolatecode

Altitude(m)

CBD incidence (%)

CBD severity (%)

I. Harenna

1
2
3
4
5
6
7

40
41
42
43

1683
1715
1656
1674
1532
1451
1420

30.0
40.0
1.0
50.0
0
0
0

8.5
15.0
1.3
13.0
0
0
0

18.6
21.2

5.4
6.6

60.0
30.0
20.0
40.0
50.0
40.0

21.0
17.5
12.5
15.0
19.0
22.5

40.0
14.1

17.9
3.7

10.0
20.0
0
0

0

6.5
3.4
0
0
0

6.0
8.9

2.0
2.9

30.0
30.0
0
0
40.0
30.0
30.0
50.0

6.3
4.5
0
0
4.2
5.0
4.5

7.8

26.3
17.7

4.0
2.8

–
–

Mean
SD
II. Bonga

1
2
3
4
5
6

50
51
52
53
54
55

1893

1872
1845
1775
1568
1663

Mean
SD
III. Berhane-Kontir

1
2
3
4
5

60
61
–
–
–

1711
1707
1185
1078
1053

Mean
SD

IV. Yayu

1
2
3
4
5
6
7
8

70
71
–
–
72
73
74
75

Mean
SD

abyssiniae) from C. arabica (Arabica) and G. xylarioides f. sp.
canephorae (F. xylarioides f. sp. canephorae) from C. canephora
(Robusta).
The pathogen exists on coffee trees in two developing
stages: Gibberella as the sexual or perfect stage producing
wind-borne ascospores and Fusarium as the asexual or imperfect stage with splash-borne conidia. Infection mostly takes
place at the imperfect stage penetrating through wounds into

the base of the stem. The fungus blocks the water supply in
the vascular system and causes a typical brown discolouration.
In the field, black to violet perithecia of the perfect stage are
formed on or beneath the bark at the base of the stem. For
the first time, Adugna et al. [31] produced perithecia of the perfect stage in vitro, when mating different isolates. The role of
ascospores for distribution of the disease and in the infection
process is not yet verified and needs to be investigated more
precisely.
During the period of assessments of the disease in
2004–2006, CWD was detected in all the indigenous coffee
field sites. The lowest percentage of infected trees was found
in Berhane-Kontir, the highest in Yayu (Table 4). Seedling
inoculation results showed that there existed significant differ-

1782
1721
1495
1475
1469
1404
1493
1675

ences among the tested accessions, and most of the coffee
accessions collected from Harenna appeared to be highly resistant to CWD with infection rates between 0% and 4.0%. Some
of the Bonga accessions had infection rates of 60–97%, Berhane-Kontir of 78–98% and Yayu of 56–98%. Seedlings of
coffee accessions possessing moderate to high resistance to
the CWD pathogen were grown, re-inoculated with the same
fungus isolate and transferred to greenhouse and field sites
for further observation.

Breeding for resistance to CLR and CBD in Kenya
Kenya is predominantly an Arabica coffee-producing country.
Coffee was introduced into Kenya by missionaries at the
beginning of the 20th century. The first plantations were established at Bura in the low lying coastal region of the country,
but due to unfavourable climatic conditions, coffee growing
was relocated to higher altitudes at Kibwezi and Kikuyu near
the capital city of Nairobi. The first variety to be introduced
and grown commercially was French Mission Coffee. Historically, cultivated Arabica coffee is derived from Bourbon and
Typica types. In the early years of coffee cultivation, the breed-


Breeding for resistance to coffee diseases

115

100
90

I Harenna
II Bonga

80

III Berhane-Kontir
IV Yayu

Frequency [%]

70

Polynomisch (Total)

60
2

y = 2E-05x + 0,0009x - 22,287

50

2

R = 0.347

40
30
20
10
0
1000

1200

1400

1600

1800

2000

Altitude [m]

Fig. 4

Incidence of CBD in the forest coffee areas of Ethiopia [24].

ing objectives of most producing countries were to select varieties combining high yield, fine beverage quality and adaptation to local growing conditions. The breeding strategy was
mainly by individual tree selections, giving rise to cultivars
SL 28, SL 34 and K 7, which are still grown commercially
today. Existing plantations of French Mission and Blue
Mountain coffee varieties are the original accessions planted
in Kenya before the selection process commenced.
C. arabica var. SL 28
The SL 28 cultivar was selected at the former Scott Laboratories
(now the National Agricultural Laboratories, NARL situated at
Kabete, Nairobi) on a single tree basis from the Tanganyika
drought resistant variety selected in Northern Tanzania in
1931. The prefix ‘‘SL’’ is an acronym for Scott Laboratories,
where the variety was selected. The name is completed by a serial
number ‘‘28’’ for the selection. The variety is suited for medium
to high altitude coffee-growing zones. It has predominantly
green shoot tips, but occasionally bronze types can be observed.
The angle of insertion of primaries is predominantly semi-erect,
but tends to become decumbent or pendant after successive
crop-bearing seasons. It has bold beans with particularly fine
liquor and is susceptible to CBD, CLR and Bacterial Blight of
Coffee (BBC), (Pseudomonas syringae pv. garcae).
C. arabica var. SL 34
SL 34 cultivar was also selected at the former Scott Laboratories from French Mission Coffee. The cultivar is adapted to
high altitude areas with good rainfall. It is mainly characterized by dark bronze shoot-tipped plants with a few greentipped strains. The laterals have a semi-erect habit, which tends

to become decumbent or drooping on older primaries. The
cultivar produces high yields of fine quality coffee, but is susceptible to CBD, CLR and BBC.

C. arabica var. K 7
K 7 cultivar was selected at Lengetet Estate in Muhoroni on
the Lake Victoria basin from the French Mission Coffee. It
is distinguished by its spreading habit on young laterals,
although older primaries tend to be decumbent or drooping.
The cultivar has characteristic medium to narrow leaves with
young shoot tips that are an intermediate bronze in colour
and shows resistance to some races of CLR, as well as partial
resistance to CBD. It is suited to lower altitudes, where CLR is
prevalent. The bean and liquor qualities are good.
Breeding objectives and selection methods
Although the above commercial varieties to a large extent met
the original breeding objectives of combining high yield with
good beverage quality and adaptation to the prevailing coffee-growing conditions, new challenges emerged that were
hitherto not addressed in the selection process. Key among
the challenges was CLR and CBD epidemics. Arabica coffee
is also known to be genetically very narrowly based due its autogamous nature [33].
Breeding for resistance to CLR took into consideration the
worldwide distribution of the disease and the multiple races of
the pathogen. In 1955, the governments of the United States of
America (USA) and Portugal established the Coffee Rust Research Centre (CIFC) in Oeiras, Portugal to coordinate CLR
research without the risk of spreading new rust races to producing countries. Resistance to CLR is inferred from Flor’s
Gene-for-Gene concept, which states that for every major
gene-conditioning resistance in the plant, there is a corresponding gene-conditioning virulence in the pathogen [34].
The resistance genes in the host are designated ‘‘SH’’ genes
while the virulence genes in the pathogen are designated ‘‘v’’.
Resistance genes SH 1–9 have been characterized and virulence

genes v 1–9 have been inferred. In a collaborative effort be-


116

H. Hindorf and C.O. Omondi

Table 4 Incidence of CWD in 2005 in the rainforest areas of
Ethiopia.
Locality

Sample sitea

Altitude (m)

CWD
incidence (%)

I. Harenna

1
2
3
4
5
6
7

1683
1715

1516
1531
1519
1476
1298

0.0
6.0
12.0
10.0
8.0
16.0
16.0

Mean
SD
II. Bonga

9.7
5.7
1
2
3
4
5

1780
1775
1568
1660

1525

Mean
SD
III. Berhane-Kontir

1707
1180
1080
1070
1053

Mean
SD
IV. Yayu

1477
1475
1404
1471
1435
1446
1493

Mean
SD
a

2.0
6.0

0
4.0
0
2.4
2.6

1
2
3
4
5
6
7

Donor Parent (D/P)

16.0
20.0
30.0
14.0
18.0
0
20.0
16.9
9.0

Number of samples: 30–50 trees/site.

tween CIFC and Arabica coffee-producing countries around
the world, several varieties resistant to rust were developed.

The most notable variety that was introduced in most countries was the Colombian Catimor, combining CLR and CBD
resistance and compact growth.
In subsequent years, management of CLR and CBD became the main subject of research and novel control strategies
combining chemical and cultural practices were developed to
manage the two diseases. Despite intensive fungicide sprays,
disease epidemics, particularly CBD, still contributed to significant economic losses, especially during prolonged cool and
wet weather conditions. Analysis of coffee production costs
further revealed that chemical control of CBD alone contributed up to 30% of the total [35]. It was further revealed that
the continuous use of some fungicides, particularly Benzimidazole compounds was found to induce the emergence of fungicide-tolerant strains [36–40]. The fungicide-tolerant strains
continued to persist in the pathogen population, even after
the fungicides were withdrawn immediately after detecting
the phenomenon [39].

Recurrent Parent (R/P)

X
F1 (50% R/P)

BC 1 ( 75% R/P)

6.0
0
8.0
10.0
8.0
6.4
3.5

1
2

3
4
5

X

X

BC 6 ( 96.8%, R/P)

Fig. 5 Schematic presentation of the backcross breeding
method.

Arising from these challenges, the breeding objective was
expanded to include the search for and the deployment of
resistance genes into existing commercial varieties that already
had good yield, beverage quality and adaptability to coffeegrowing conditions, using the backcross breeding method
(Fig. 5). In Kenya, the breeding programme was initiated in
1971 as a bilateral partnership between the Kenya Government and the Netherlands Government. Realizing that the
commercial cultivars grown in Kenya were mostly susceptible
and that there was very little variability within the Arabica coffee germplasm in Kenya, an aggressive campaign to introduce
accessions and landraces from other coffee-growing countries
in Latin and Central America and particularly from the centre
of origin of Arabica coffee in the southwest highlands of Ethiopia, was launched. The resulting genetic pool, comprising of
the world coffee germplasm collection and the introductions
of the 1964 FAO coffee mission to Ethiopia, provided the
source of genetic variation from which to select for resistant
genotypes [41]. Inheritance studies using 11 Arabica coffee
varieties varying in CBD resistance revealed three major genes
on separate loci [42]. The highly resistant variety, Rume Sudan

originating from the Boma Plateau in southern Sudan, carries
the dominant R- and the recessive k-gene. The R-locus has
multiple alleles with R1R1 in Rume Sudan and R2R2 in Pretoria, which also carries the recessive k-gene. The moderately
resistant variety K 7 carries the recessive k-gene. Clone 1349/
269 of the variety Hibrido de Timor and its hybrid derivative
Catimor carries one gene for CBD resistance on the T-locus
with intermediate gene action.
A gene deployment strategy that would combine two or
more resistance genes in the same plant and create variability
through gene recombination in segregating populations arising
from single, double, three way and multiple crosses was initiated. The resulting crosses were backcrossed to the susceptible
commercial varieties to restore good yield, fine beverage qual-


Breeding for resistance to coffee diseases
ity and adaptability to local growing conditions while selecting
for resistance in the resultant progeny as inherited from resistant donor parents (Fig. 5).
The breeding programme got a boost when the Catimor
variety was introduced from Colombia. It was found to be
resistant to CBD on the T-locus and to all the races of the
CLR pathogen found in Kenya. The variety was also compact
in growth, which presented an opportunity for high density
planting. However, it could not be released as a commercial
variety in Kenya, because the genetic base for CBD resistance
was narrow (one gene) and the beverage quality required to
be improved to the standard of SL 28, SL 34 and K 7. A strategy
was adopted to use the Catimor variety as mother parent and
the progeny of the backcross breeding programme cited above
as the male parent in a hybrid seed production scheme. A variety combining the attributes of the Catimor variety and the
backcross progeny was released in 1985 and named ‘‘Ruiru 11’’.

C. arabica var. Ruiru 11
The variety name has the prefix ‘‘Ruiru’’ referring to the location of the Kenyan Coffee Research Station where the variety
was developed. The name is completed by an additional two
code numbers, ‘‘11’’. The first code number denotes the sequence of release, in this case the first release, and the second
number defines the type of variety as a one-way cross between
two designated parent populations. The variety is not only
resistant to CBD and CLR but is also compact in growth,
allowing farmers to intensify the production per unit of land,
especially in high potential areas, where the human population
is high and coffee is in competition with other crops and farm
enterprises required for food security and income. Ruiru 11 is
planted at a density of 2500–3300 trees/ha compared to 1300
trees/ha for traditional varieties. This translates into a higher
production per unit area of land. The variety comes into production earlier, hence earlier realization of benefits for farmers.
The development of Ruiru 11 also took into consideration the
importance of quality as a major marketing parameter. Since
the quality of the traditional varieties was already popular
among consumers of Kenyan coffee, Ruiru 11 was developed
with quality attributes similar to the traditional varieties, SL
28, SL 34 and K 7.
Despite the successful performance of the Ruiru 11 variety,
the major drawback has been the availability of adequate seeds
to meet the high demand of growers both locally and in the region. As a hybrid variety, seed multiplication involves artificial
cross pollination between the male and female parents. Noting
that there has been no male sterility documented in coffee, artificial cross pollination requires manual emasculation of the female plants and pollination by the male plants. This is a labour
intensive process that has continued to limit the amount of
seeds that can be produced. Following the large scale cultivation of Ruiru 11 over several years, it has also been necessary
to study the variation in the CBD pathogen. There has been no
evidence of breakdown of resistance but differences in the
aggressiveness of isolates are sometimes prominent [43].

CWD has not been reported in Kenya despite its close
proximity to Uganda where the disease has ravaged Robusta
plantations, because Kenya is predominantly an Arabica
coffee-producing country. Ethiopia, which shares its southern
border with Kenya, is the only country, where CWD has been

117
detected on Arabica coffee, but it is believed that the arid
Northern province of Kenya provides a buffer zone, hindering
the spread of the disease into Kenya’s coffee plantations.
Breeding for resistance to CWD has therefore gained prominence in Uganda and Ethiopia, where the main focus is selection within the local landraces.
Recent progress in the variety improvement and development of a
true breeding resistant variety
A breeding approach to develop a true breeding variety is
currently in progress in Kenya. The variety has been entered
into a pre-release adaptation trial. It was developed from
individual tree selections of backcross progeny involving
SL 4, N 39, Hibrido de Timor and Rume Sudan as the donor varieties and cvs. SL 28, SL 34 and K 7 as the recurrent
parents. In this method, the best individuals within the best
families were selected solely on the basis of their phenotypic
values (within the family selection method). The strategy involved simultaneous selection for the important traits, but
independent rejection of all the individuals that failed to
meet the required standard for any one of the traits under
improvement (independent culling level). The performance
of cultivar Ruiru 11 was used as a standard check for
discriminating against inferior lines when selecting for resistance to CBD and CLR, yield and quality. The variety SL
28 was also used as a standard when selecting for yield
and quality.
The variety is a composite of five crosses (cross 8, cross 22,
cross 23, cross 27 and cross 30) that are tall in stature, the distinctive features being true breeding, resistance to CBD and

CLR. It is a high-yielding variety with good bean and liquor
quality that is comparable to Ruiru 11 and SL 28, suited for
all coffee agro-ecological zones in Kenya and has a conical
shape with a horizontal but occasionally erect branching habit,
which tends to become semi-drooping or drooping after successive crop-bearing seasons. The young leaves have medium
anthocyanin colouration giving a bronze colour, occasionally
absent or weak, giving a green-bronze colouration. Yield data
indicate that the crosses are better than or comparable to the
standard check varieties, Ruiru 11 and SL 28 (Table 5). Disease assessment data revealed that CBD infections were significantly higher in the susceptible SL 28 than in the treatments
and resistant Ruiru 11 control (Table 5). CLR infection
showed clear variations between the susceptible SL 28 on the
one hand and the resistant crosses and Ruiru 11 control on
the other. It is important to note that resistance among the
crosses was not significantly different from the resistant Ruiru
11.
Molecular approaches to coffee breeding
Efficient and reliable disease screening methods are required
for a successful variety development programme. Molecular
markers linked to resistance provide the potential to screen
for resistance in a large population of plants at any stage
of plant development. Where several genes confer resistance,
markers have the advantage over morphological assessments,
because plants carrying multiple resistance (broad-based
resistance) can easily be differentiated from those carrying a
single gene (narrow-based resistance). Attempts have been


118

H. Hindorf and C.O. Omondi

Table 5 Mean yield performance and disease score of the five test genotypes and control varieties (Ruiru 11 and SL 28) at Tatu Estate
in Ruiru/Kenya.
Treatment

Cross 08
Cross 22
Cross 23
Cross 27
Cross 30
Ruiru 11
SL 28

Mean yield of clean
coffee (g/tree)

Disease score
28.6.2007

1718.66BC
1045.93D
1966.36AB
1292.80CD
2539.03A
1877.40BC

26.7.2007

17.9.2007

Mean CBD

Mean CLR

Mean CLR

Mean CLR

0.12A
0.3267A
0.0967A
0.1433A
0.0333A
0.1133A
1.07B

0.1833AB
0.45C
0A
0.35BC
0.3BC
0.1333AB
1.0167D

0.2333AB
0.4667B
0.15AB
0.4333B
0A
0.2167AB

1.4333C

0.2667A
0.4167A
0.1333A
0.3833A
0A
0.15A
2.1667B

Note: Means followed by a common letter(s) are not significantly different according to Duncan’s Multiple Range Test (P = 0.05).

F 2 Progeny – Tree Numbers
M

P1

P2

29

41 69

95

111

6

22

26

82

94

9

28

51

70

112

Blank

200 bp

150 bp

180 bp

Primer Dimer

M = 100 bp ladder, P1 =SL 28, P2 = Rume Sudan.

Fig. 6

SSR polymorphism using primer M2.

made to screen for DNA markers linked to CBD resistance in
Catimor and Rume Sudan coffee varieties [44,45]. For instance, DNA was extracted from an F2-mapping population
derived from Rume Sudan · SL 28 using the method of Diniz
et al. [46]. The DNA was subjected to microsatellite analysis
using primer M 24 that had forward and reverse sequences
50 GGCTCG AGATATCTGTTTAG30 and 50 TTTAATGGGCATAGGG TCC30 , respectively, and repeat motif as
(CA)15(CG)4CA [47]. The primer detected polymorphism at
three different levels (Fig. 6). The susceptible parent, SL 28,
amplified a fragment of 150 bp size, while the resistant parent, Rume Sudan, amplified a fragment of about 180 bp.
These fragments were also evident in some F2 progeny.
The third category of fragments appeared in pairs and was
mainly observed in the F2 plants. This category is believed
to comprise the heterozygotes. Omondi et al. [45] concluded
that the observed SSR polymorphism is consistent with major
gene inheritance. The resistant Rume Sudan variety is known

to carry a dominant gene for CBD resistance on the R-locus
[42]. More investigations are still necessary to establish with
precision the trait that co-segregates with the observed
DNA bands so as to conclude that the bands that represent
markers for a specific target trait. Efforts have now been directed to determine the genotypes of individual plants constituting of the mapping population using the hypocotyls
inoculation test. The potential use of the bands as markers
for selection will depend on their potential to co-segregate
with resistance/susceptibility to CBD.
Conclusions
Ethiopia is the centre of origin of C. arabica and there exists an
immense opportunity to develop and use resistant varieties to

manage diseases. The existence of a tremendous diversity
of different characteristics was observed in Arabica coffee


Breeding for resistance to coffee diseases
[4,47,48,51]. This observation was recently ascertained by
molecular analysis [49,50]. Various investigations demonstrated the presence of resistance to CLR, CBD and CWD
in Arabica coffee collected from Ethiopia [4,7,24,50].
Tremendous achievements have been realized by breeding
coffee varieties resistant to fungal diseases. Resistant varieties
have the potential to reduce the cost of production and
offer environmentally better disease management approaches.
Novel methods of selection that can reduce the time taken for
variety development are being explored through molecular
marker approaches.
Acknowledgements
The authors would like to thank their partners for carrying
out most of the research work: Arega Zeru (CBD), Chala Jefuka (CLR), Dr. Gima Adugna (CWD) and the Jimma Agricultural Research Centre (JARC) in Ethiopia for providing
working facilities in the laboratory and field sites. The Centre
for Development Research (ZEF) of the University of Bonn
coordinated the large scale project ‘‘Conservation of wild coffee in the montane rainforests of Ethiopia (CoCE)’’. Financial support was granted by the German Ministry of
Research and Education (BMBF) and the German Foreign
Exchange Service (DAAD). The authors further acknowledge
the breeding work done by Dr. H. van der Vossen and Dr.
D.J. Walyaro at the Coffee Research Foundation in Kenya,
which has formed a major basis of the review on breeding
progress.
References
[1] Hindorf H. Current diseases of Coffea arabica and C. canephora
in East Africa causing crop losses. Meded Fac Landbouwwet

Univ Gent 1998;63:861–5.
[2] Waller JM. Coffee rust epidemiology and control. Crop Prot
1982;1(4):385–404.
[3] Sylvain PG. Ethiopian coffee – its significance to world coffee
problems. Econ Bot 1958;12(2):111–39.
[4] Van der Graaff NA. Selection of Arabica coffee types resistant
to coffee berry disease in Ethiopia. Wageningen: Mededelingen
Landbouwhogeschool; 1981.
[5] Ritschel A. Monograph of the genus Hemileia (Uredinales). Borntraeger; 2005.
[6] Wondimu M, Hulluka M, Rodrigues CJJ. Distribution of races
of Hemileia vastatrix B & Br and physiologic resistance groups
of Coffea arabica L. in Ethiopia. Ethiop J Agric Sci
1987;9(1):25–38.
[7] Adugna G, Jefuka C, Teferei D, Zeru A. Multiple resistances to
Coffee Berry Disease, Coffee Wilt and Leaf Rust in Coffea
arabica populations of Ethiopia. In: XXII ASIC, Campinas, SP,
Brazil; 2008.
[8] McDonald J. A preliminary account of a disease of green coffee
berries in Kenya colony. Trans Br Mycol Soc 1926;11(1–
2):145–54.
[9] Waller JM, Bigger M, Hillocks RJ. Coffee pests, diseases and
their management. 1st ed. CABI; 2007.
[10] Rayner RW. Coffee berry disease: a survey of investigations
carried out up to 1950. E Afr Agric J 1952;17:130–58.
[11] Noack D. Die Krankheiten des Kaffeebaumes in Brasilien. Z
Pflanzenkr Pflanzenpathol 1901;11:202.
[12] Waller JM, Bridge PD, Black R, Hakiza G. Characterisation of
the coffee berry disease pathogen, Colletotrichum kahawae sp.
nov. Mycol Res 1993;97(8):989–94.

119
[13] Gibbs JN. Inoculum sources for coffee berry disease. Ann Appl
Biol 1969;64:515–22.
[14] Hindorf H. Colletotrichum spp. isolated from Coffea arabica L.
in Kenya. Z PflKrankh PflSchutz 1970;77:328–31.
[15] Hindorf H. Colletotrichum-Arten aus den Kaffeeanbaugebieten
von Kiambu in Kenia. Z PflKrankh PflSchutz 1974;81:108–13.
[16] Hindorf H. Colletotrichum – population auf Coffea arabica L. in
Kenia: I. Eine Methode zur systematischen Trennung von
Pilzpopulationen. Phytopathol Z 1973;77:97–116.
[17] Hindorf H. Colletotrichum – population auf Coffea arabica L. in
Kenya: II. Qualitative und quantitative Unterschiede in der
Colletotrichum – population. Phytopathol Z 1973;77:216–34.
[18] Hindorf H. Colletotrichum – populationen auf Coffea arabica L.
in Kenia: III. Verbreitung von Colletotrichum-Arten auf den
einzelnen Organen des Kaffeestrauches. Phytopathol Z
1973;77:324–38.
[19] Hindorf H. Colletotrichum occurring on Coffea arabica: a
review. J Coffee Res 1975;5(3/3):43–56.
[20] Firman ID, Waller JM. Coffee berry disease and other
Colletotrichum diseases of coffee. 4th ed. Kew, Surrey,
England: Commonwealth Mycological Institute; 1977.
[21] Mogk M. Report on the coffee disease survey in
Ethiopia. Addis Ababa/ET: IAR; 1971.
[22] Biratu T. Studies on Colletotrichum populations of Coffea arabica
L. in Ethiopia and evaluation of reactions of coffee germplasm.
Ph.D. thesis, University of Bonn/Germany; 1995.
[23] Mogk M, Hindorf H. Verluste durch die KaffeekirschenKrankheit (Colletotrichum coffeanum Noack) in verschiedenen
Stadien der Kirschen-Entwicklung. Z PflKrankh PflSchutz
1975;82:193–200.

[24] Zeru A, Assefa F, Adugna G, Hindorf H. Occurrence of fungal
diseases of Coffea arabica L. in montane rainforests of Ethiopia.
J Appl Bot Food Qual 2009;82(2):148–51.
[25] Lejeune JBH. Rapport au gouvernement imperial d’Ethiopie sur
la production cafe´ie`re. Rome: FAO; 1958.
[26] Kranz J, Mogk M. Gibberella xylarioides (Heim et Saccas) on
Arabica coffee in Ethiopia. Phytopathol Z 1973;78:365–6.
[27] Adugna G, Hulluka M, Hindorf H. Incidence of tracheomycosis,
Gibberella xylarioides (Fusarium xylarioides), on Arabica coffee in
Ethiopia. Z PflKrankh PflSchutz 2001;108(2):136–42.
[28] Adugna G. Diversity in pathogenicity and genetics of Gibberella
xylarioides (Fusarium xylarioides) populations and resistance of
Coffea spp. in Ethiopia. Ph.D. thesis, University of Bonn/
Germany; 2004.
[29] Adugna G, Hindorf H, Steiner U, Nirenberg HI, Dehne HW,
Schellander K. Genetic diversity in the coffee wilt pathogen
(Gibberella xylarioides) populations: differentiation by host
specialization and RAPD analysis. Z PflKrankh PflSchutz
2005;112(2):134–45.
[30] Adugna G, Flood J, Hindorf H, Bieysse D, Sarah S, Rutherford
M. Tracheomycosis (Gibberella xylarioides). A menace to world
coffee production: evidenced by cross inoculation of historical
and current strains of the pathogen. In: XXI ASIC, Montpellier/
F; 2006.
[31] Adugna G, Hindorf H, Steiner U, Dehne HW. Mating test and
in vitro production of perithecia by the coffee wilt pathogen,
Gibberella xylarioides (Fusarium xylarioides). Ethiop J Biol Sci
2007;6(1):63–75.
[32] Adugna G, Million A, Hindorf H, Zeru A, Jefuka C. Coffee
Wilt Disease in Ethiopia. In: Flood J, editor. Coffee Wilt

Disease. CAB International; 2010. p. 43–61.
[33] Lashermes P, Combes MC, Robert J, Trouslot P, D’Hont A,
Anthony F, et al. Molecular characterisation and origin of the
Coffea arabica L. genome. Mol Gen Genet 1999;261(2):259–66.
[34] Flor HH. Current status of the gene for gene hypothesis. Annu
Rev Phytopathol 1971;9:275–96.
[35] Nyoro JK, Sprey LH. Introducing Ruiru 11 to estates and small
holders. Kenya Coffee 1986;51:7–28.


120
[36] Cook RTA, Pereira JL. Strains of Colletotrichum coffeanum, the
causal agent of coffee berry disease tolerant to benzimidazole
compounds in Kenya. Ann Appl Biol 1976;83:365–79.
[37] Okioga DM. Occurrence of strains of Colletotrichum coffeanum
resistant to methyl benzimidazole-2-ylcarbamate(carbendazim)
and chemically similar compounds. Ann Appl Biol
1976;84:21–30.
[38] Javed ZUR. Benomyl tolerance in Colletotrichum coffeanum, the
cause of coffee berry disease. Kenya Coffee 1980;45:87–91.
[39] King’ori PN, Masaba DM. Distribution and persistence of
benomyl resistant populations of Colletotrichum coffeanum in
coffee. Kenya Coffee 1991;56:1071–4.
[40] Mwang’ombe AW, Mukunya DM, Gathuru EM. Some
mechanisms implicated in the survival of benomyl tolerant
strains of Colletotrichum coffeanum Noack, casual agent of
coffee berry disease. Dis Innovation 1992;4(2):109–15.
[41] Omondi CO. Resistance to coffee berry disease in Arabica coffee
variety Ruiru-11. Plant Breed 1994;112(3):256–9.
[42] van der Vossen HAM, Walyaro DJ. Breeding for resistance to

coffee berry disease in Coffea arabica L.: II. Inheritance of the
resistance. Euphytica 1980;29(3):777–91.
[43] Omondi CO, Ayiecho PO, Mwang’ombe AW, Hindorf H.
Resistance of Coffea arabica cv. Ruiru 11 tested with different
isolates of Colletotrichum kahawae, the causal agent of coffee
berry disease. Euphytica 2001;121(1):19–24.
[44] Gichuru EK, Agwanda CO, Combes MC, Mutitu EW, Ngugi
ECK, Bertrand B. Identification of molecular markers linked to
a gene conferring resistance to coffee berry disease
(Colletotrichum kahawae) in Coffea arabica. Plant Pathol
2008;57(6):1117–24.

H. Hindorf and C.O. Omondi
[45] Omondi CO, Gichuru EK, Combes MC, Lashermes P. SSR
polymorphism in breeding populations of Arabica coffee with
varying reactions to coffee berry disease. In: Plant and animal
genomes XVII conference. San Diego, CA: Town & Country
Convention Center; 2009. p. 44–8.
[46] Diniz L, Sakiyama N, Lashermes P, Caixeta E, Oliveira A,
Zambolim E. Analysis of AFLP markers associated to the Mex1 resistance locus in Icatu progenies. Crop Breed Appl
Biotechnol 2005;5:387–93.
[47] Combes MC, Andrzejewski S, Anthony F, Bertrand B, Rovelli
P, Graziosi G. Characterization of microsatellite loci in Coffea
arabica and related coffee species. Mol Ecol 2000;9(8):1178–80.
[48] Bayetta B, Behailu A, Gebramu T. Testing of f5 lines of Catimor
Hybrid (Cattura vermehelo · Hybrido de Timor) for adaptation.
In: Proceedings of the eighth annual conference of crop science
of Ethiopia (CSSE), Addis Ababa/ET; 1999. p. 9–15.
[49] Paulos D, Demel T. Coffee production systems in Ethiopia. In:
Proceedings of the workshop on control of coffee berry disease

in Ethiopia, 13–15, 1999, Addis Ababa, Ethiopia/ET; 2000. p.
99–107.
[50] Aga E, Bryngelsson T, Bekele E, Salomon B. Genetic diversity
of forest Arabica coffee (Coffea arabica L.) in Ethiopia as
revealed by random amplified polymorphic DNA (RAPD)
analysis. Hereditas 2003;138(1):36–46.
[51] Kassahun T, Tamiru O, Govers K, Endashaw B, Borsch T.
Genetic diversity and population structure of wild Coffea
arabica populations in Ethiopia using molecular markers. In:
Girma A, Bayetta B, Tesfaye S, Endale T, Taye K, editors.
Coffee diversity and knowledge. Addis Ababa/ET: EIAR;
2008. p. 35–44.